Differential Proteomic Analysis of the Bacillus anthracis Secretome: Distinct Plasmid and Chromosome CO₂-Dependent Cross Talk Mechanisms Modulate Extracellular Proteolytic Activities

Abstract

The secretomes of a virulent Bacillus anthracis strain and of avirulent strains (cured of the virulence plasmids pXO1 and pXO2), cultured in rich and minimal media, were studied by a comparative proteomic approach. More than 400 protein spots, representing the products of 64 genes, were identified, and a unique pattern of protein relative abundance with respect to the presence of the virulence plasmids was revealed. In minimal medium under high CO₂ tension, conditions considered to simulate those encountered in the host, the presence of the plasmids leads to enhanced expression of 12 chromosome-carried genes (10 of which could not be detected in the absence of the plasmids) in addition to expression of 5 pXO1-encoded proteins. Furthermore, under these conditions, the presence of the pXO1 and pXO2 plasmids leads to the repression of 14 chromosomal genes. On the other hand, in minimal aerobic medium not supplemented with CO₂, the virulent and avirulent B. anthracis strains manifest very similar protein signatures, and most strikingly, two proteins (the metalloproteases InhA1 and NprB, orthologs of gene products attributed to the Bacillus cereus group PlcR regulon) represent over 90% of the total secretome. Interestingly, of the 64 identified gene products, at least 31 harbor features characteristic of virulence determinants (such as toxins, proteases, nucleotidases, sulfatases, transporters, and detoxification factors), 22 of which are differentially regulated in a plasmid-dependent manner. The nature and the expression patterns of proteins in the various secretomes suggest that distinct CO₂-responsive chromosome- and plasmid-encoded regulatory factors modulate the secretion of potential novel virulence factors, most of which are associated with extracellular proteolytic activities.

Bacillus anthracis is a gram-positive spore-forming bacterium that is the etiological agent of anthrax, a lethal disease sporadically affecting humans and animals, in particular herbivores. In its most severe manifestation, B. anthracis infection is initiated by inhalation of spores, which are taken up by alveolar macrophages and germinate into fast-dividing vegetative cells which secrete toxins and virulence factors during growth (81, 99). If untreated by prompt antibiotic administration, the bacteria invade the bloodstream, resulting in massive bacteremia and consequently generalized systemic failure and death. B. anthracis is considered to represent a potential biothreat agent, owing to the severity of the anthrax disease, the ease of respiratory contamination, and the perpetual environmental stability of the infective spores. The recent deliberate dissemination of B. anthracis (15) accelerated the efforts to identify new B. anthracis virulence-related determinants for the design of novel diagnostic, preventive, and/or therapeutic strategies.

Fully virulent B. anthracis strains harbor two native plasmids, pXO1 and pXO2, which encode critical pathogenicity factors. The absence of either one of the two plasmids results in a pronounced attenuation of B. anthracis virulence. The pXO2 plasmid encodes proteins involved in the biosynthesis of the poly-d-glutamic acid capsule, which may inhibit phagocytosis of bacteria during infection; pXO1 encodes the three toxin components protective antigen (PA), lethal factor (LF) (a zinc-dependent metalloprotease which proteolytically inactivates protein kinase kinases 1 and 2), and edema factor (EF) (a calmodulin-dependent adenylate cyclase), which form two binary toxins, lethal toxin and edema toxin. PA, the common component of both toxins, is not toxic by itself, yet it plays the central role of binding a specific receptor on the host cells and translocating LF and EF into the cytosol of infected cells, where they exert their detrimental activities. Anthrax is acknowledged as a toxinogenic disease, owing to the lethality of pure toxin preparations (77); on the other hand, additional B. anthracis secreted proteins are most probably involved in the onset and course of the disease and in survival of the bacteria in the host.

The regulatory circuits governing the virulence of B. anthracis are still to be fully deciphered, yet certain observations suggest that the virulence of the bacteria entails cross talk mechanisms which link expression of plasmid-encoded and chromosomally encoded genes. The regulatory AtxA protein, encoded by pXO1, is essential for expression of the toxin and capsule synthesis genes in vivo (a situation which can be mimicked by growing the bacteria in minimal medium under high bicarbonate-CO₂ conditions) (75). Two additional regulatory proteins, AcpA and AcpB, encoded by pXO2, were suggested to act downstream of AtxA and to affect capsule synthesis (35, 36). AtxA was found also to influence expression of chromosomal genes, either directly or via AcpA and AcpB. In addition, the protein AbrB, which is a chromosomally encoded transition state regulator, was suggested to negatively control the activity of the toxin gene promoters (123, 131) via AtxA.

Secreted proteins include factors involved in pathogenicity, in particular in gram-positive bacteria (79). Such proteins may serve as possible targets for diagnostic purposes and/or therapeutic intervention. Bacteria of the Bacillus cereus phylogenetic group (B. cereus and Bacillus thuringiensis), to which B. anthracis belongs, secrete a diversity of factors that are essential for virulence, including toxins, hemolysins, proteases, and lecithinases. Notably, in these bacteria the secretion of certain virulence factors is regulated by a pleiotropic regulator, PlcR (2, 82, 87, 110), which is inactive in B. anthracis (2). It has been suggested that the evolutionary inactivation of the PlcR regulon in B. anthracis was due to incompatibility with the AtxA-controlled regulon and reflects the fact that the PlcR target genes are not essential for anthrax pathogenicity (96). Several studies have postulated that secreted proteases, other than those belonging to the silenced PlcR regulon, are responsible for some clinical manifestations of anthrax (1, 9, 117, 140). Such proteases could damage host tissues, interfere with immune effectors of the host, and/or provide nutrients for bacterial survival (103). Some chromosomally encoded B. anthracis extracellular proteases were suggested to be controlled by Cot43, a novel regulatory gene encoded by pXO1 (9).

The availability of the B. anthracis genomic DNA sequence (109, 121) paved the way for high-throughput genomic, transcriptomic, and proteomic analyses of B. anthracis (6, 7, 8, 17, 25, 41, 50, 64, 78, 85, 121, 144) in an effort to elucidate pathogenicity mechanisms by identification of novel virulence factors or in search for specific therapeutic and/or diagnostic targets. Indeed, bioinformatic surveys of the B. anthracis genome (7, 8, 121) suggested that proteins other than PA, LF, and EF, may participate in anthrax pathogenesis. Furthermore, many B. anthracis open reading frames (ORFs) encode potentially secreted or membrane-bound proteins exhibiting homology to known virulence factors from other bacteria (7, 8).

A preliminary proteomic study carried out in our laboratory examined membrane proteins prepared from a nonvirulent B. anthracis strain and led to the recognition of a number of immunodominant exposed proteins (8, 25). Here, we document an extended proteomic study, focusing on B. anthracis secreted proteins, which expands the data set of expressed B. anthracis proteins from both virulent and nonvirulent strains (6, 25, 41, 50, 64, 78, 85, 144). Based on identification of more than 400 two-dimensional electrophoresis (2-DE)-separated protein spots, we report the expression of 64 proteins which represent the most abundant B. anthracis secreted proteins, many of which resemble factors involved in the virulence of other pathogens. Comparison of the relative abundances of proteins in pXO1- and pXO2-containing and plasmid-cured strains reveals about 30 ORFs which are either preferentially expressed or repressed in the virulent B. anthracis strain under conditions which are considered to simulate those encountered within the mammalian host. The pattern of expression of these specific proteins demonstrates that B. anthracis possesses distinct regulatory pathways which involve plasmid- or chromosome-encoded CO₂-inducible responsive factors.

MATERIALS AND METHODS

Bacterial cultures and sample preparation for 2-DE.

The B. anthracis strains used in this study are: the fully virulent Vollum strain (pXO1⁺ pXO2⁺) and the attenuated strains ΔVollum (pXO1⁻ pXO2⁻) and Δ14185 (pXO1⁻ pXO2⁻). Strain Δ14185 is a pXO1-deleted derivative of the nonproteolytic vaccine strain V770-NP1-R (148) (ATCC 14185).

Cells were cultured in either FAG medium (27), brain heart infusion (BHI) (Difco/Becton Dickinson, MD), or NBY medium (containing 0.8% [w/vol] nutrient broth [Difco], 0.3% yeast extract [Difco], and 0.5% glucose) for up to 24 h at 37°C with vigorous agitation. The rich FAG and BHI media were selected following a preliminary pilot study in which different media routinely used for B. anthracis cultures were examined for their ability to support maximal cell yield and to promote expression of a rich complementary repertoire of extracellular proteins (data not shown). For identification of CO₂-induced proteins, cells were grown at 37°C in NBY supplemented with 0.9% NaHCO₃ in hermetically sealed filled flasks without agitation (referred to hereafter as NBY-CO₂). Under these conditions, which we define as semiaerobic, no aeration of the culture occurs during growth, except for the small amount of air present at the onset of the culture. After 12 h of growth, the bacterial cultures typically reach optical densities at 660 nm of approximately 15, 12, and 6 in FAG, BHI, and NBY, respectively. In NBY-CO₂, an optical density at 660 nm of 1 to 2 is observed after 12 h. In all cases, no visible change in optical density was observed between 12 and 36 h of culture. The low-nutrient-content (compared to FAG or BHI) NBY-CO₂ medium promotes very efficient toxin production and capsule synthesis (which can be visualized by negative staining using India ink [Becton Dickinson, MD]).

Secreted proteins were collected from the culture-conditioned media essentially as described by Antelman and coworkers (4). In brief, cultures were centrifuged for removal of cells and subjected to filtration using 0.22-μm filters. One hundred milliliters of FAG or BHI conditioned medium or 200 ml of NBY medium was incubated in the cold overnight in the presence of 10% trichloroacetic acid and then precipitated by centrifugation for 30 min in a Sorvall S34 instrument (12,000 rpm). Pellets of trichloroacetic acid-precipitated proteins were washed four times in a large volume of 96% ethanol and then resuspended by scraping and extensive pipetting in 5 ml of an isoelectric focusing (IEF) sample solution composed of 8 M urea, 4% (wt/vol) 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS), 40 mM Tris, 2% dithiothreitol (DTT), and 0.2 (wt/vol) Bio-Lyte 3/10 (Bio-Rad). The clear solution contained approximately 0.3 mg/ml protein.

2-DE separation of proteins and spot quantitation.

The secreted protein mixture (100 μg/run, representing approximately 7 ml of FAG or BHI culture medium or 15 ml of NBY medium) was resolved first by IEF on pH 3 to 10 (nonlinear) ready-made, 17-cm, immobilized pH gradient strips (Immobiline DryStrips; Pharmacia), applied to a Protean IEF cell (Bio-Rad). IEF was carried out at 10,000 V to a total of 50,000 V-h, initiated by a slow step at 250 V for 30 min. Strips were then processed for the second-dimension separation by a 10-min incubation in 6 M urea, 2% sodium dodecyl sulfate (SDS), 0.375 M Tris-HCl (pH 8.8), 20% glycerol, 2% (wt/vol) DTT, followed by a 10-min incubation in a similar solution in which the DTT was replaced by 2% iodoacetamide. Strips were applied to 12.5% SDS-polyacrylamide gels, and electrophoresis was carried out on an Ettan DALT II System (Pharmacia). Gels were stained with Coomassie blue G-250 (Bio-safe Coomassie; Bio-Rad) and spots detected and analyzed by scanning on a GS-800 calibrated densitometer assisted by the PDQuest 2-D software (Bio-Rad). In each case, at least two independently obtained secretomes representing the same biological sample were evaluated by 2-DE, and for each sample, at least three gels were used. Apparent induction or repression of a particular protein was considered to have occurred if the relative intensities of its respective pairwise spots in the compared gels exhibited at least a fivefold difference. Reference protein spots with essentially identical intensities were used for quantitative comparisons. It is conceivable that the use of a more sensitive staining procedure could detect additional protein spots, yet the Coomassie blue staining (approximate detection level of 50 ng protein/spot) enabled visualization of the most abundant protein species as well as the major changes associated with the culture conditions.

In-gel trypsin digestion and MALDI-TOF identification of proteins.

Protein spots were cut from 2-DE gels, destained for 1 h in 30% CH₃CN, 50 mM NH₄HCO₃, and subjected to in-gel overnight digestion with 10 μl of 6.25-μg/ml trypsin (Promega). Gel fragments were subjected to two rounds of vacuum drying and resuspension in double-distilled water, and then peptide elutions were carried out first in 1% trifluoroacetic acid (TFA) (20 min, room temperature) and then in 50% CH₃CN (20 min, room temperature). Finally, the eluted tryptic peptides were dried under vacuum and resuspended in 10 ml of 25% CH₃CN, 0.1% TFA. Two microliters was mixed with an equal volume of 10-mg/ml α-cyano aqueous solution containing 50% ethanol, 25% CH₃CN, and 0.1% TFA; applied to a matrix-assisted laser desorption ionization-time-of-flight (MALDI-TOF) target; and allowed to crystallize in a vacuum oven at 60°C. Mass spectra were acquired on a Micromass TofSpec 2E instrument in positive ion reflectron mode, using a source voltage of 20,000, a pulse voltage of 2,600 to 3,000, and a laser intensity of 20%. External calibration using standard peptides was applied. Tryptic fragments originating from trypsin were used as internal standards for correction of mass spectra. MALDI-TOF spectra were processed using the MassLynx mass spectrometry software (Micromass) and compared to hypothetical tryptic digestion fragments of all ORFs in the B. anthracis Ames ancestor strain genome DNA sequence by two alternative modalities: (i) online, using the ProFound search engine at http://bioinformatics.genomicsolutions.com/knexus.html (confidence of identification was based on a Z score of above 2), or (ii) using an in-house-generated identification program described previously (8, 25). Throughout this paper, chromosomal ORFs are identified according to the NCBI locus tag identifier of the B. anthracis Ames ancestor chromosome and plasmid-located ORFs are identified by their ORF identifier according to Okinaka and coworkers (109). Accession numbers of the Ames ancestor genome are NC_007530 for the chromosome, NC007322 for pXO1, and NC007323 for pXO2 (www.ncbi.nlm.nih.gov/genomes/MICROBES/anthracis.html). Identification of proteins was based on peptide coverage of more than 30% and peptide mass deviation between observed and calculated values of less than 100 ppm. In cases in which the two-dimensional electrophoretic migration of proteins indicated a molecular weight lower than predicted, such as for proteins which undergo proteolytic processing, their MALDI-TOF tryptic-digest spectra served for determining fragments present in the mature forms, by establishing the positions of the detected peptides on the primary amino acid sequences of the proteins. For the sake of simplicity, protein spots representing the products of the same ORF were assigned the same spot number throughout this study.

Sequence analysis of B. anthracis ORFs.

Throughout this study, functional annotation of the B. anthracis Ames strain chromosomal or plasmid ORFs was according to Ariel and coworkers (7, 8). Putative localization of the product of each ORF was assigned based on integration of results of analyses for the presence of signal peptides, lipoprotein-anchoring signals, transmembrane segments, gram-positive-specific anchoring motifs, and peptidoglycan-anchoring domains, as previously described (8). The amino acid sequences of all proteins identified were subjected to BLAST analysis against all available bacterial genomes of the NCBI data bank for identification of ortholog and paralog genes. Potential export signal peptides were identified using the SignalP (version 3.1) server (http://www.cbs.dtu.dk); in cases when signal sequences could not be predicted by the computerized analysis, the N terminus of the respective protein sequence was inspected individually for the presence of typical secretion signal N and H regions as well as consensus cleavage sites (135, 136, 139). In one case (protein BA1197), the amino acid sequence of the protein lacks a characteristic signal sequence. Reexamination of the genomic DNA sequence upstream of the putative translation starting point revealed a correct type II signal peptidase (SPase) cleavage signal in an alternative reading frame; this signal is 100% homologous to that present in the B. cereus ortholog. We assume that a DNA sequence mistake resulting in a frameshift in the deduced protein sequence occurred approximately 40 base pairs upstream of the presently assigned translation initiation point.

Putative conserved functional domains were identified by sequence analysis using the CDD database on the NCBI protein analysis server (91). The predicted ligand specificities of all solute-binding domains (SBPs) of ABC transporters detected in this study are according to the TransportDB classification (http://www.membranetransport.org/other).

Zymography.

In-gel proteolytic activity assays were carried out essentially as described by Caballero and coworkers (20). Twenty-five-microliter bacterial conditioned medium samples were electrophoresed under nonreducing conditions using 10% SDS-polyacrylamide gel electrophoresis with 0.1% gelatin, casein, or purified PA (27). The gels were washed three times in 2.5% Triton X-100 and incubated for 24 h at 37°C in either casein substrate buffer (50 mM Tris [pH 7.6], 1 μM ZnCl₂, 100 mM NaCl) or gelatin/PA substrate buffer (50 mM Tris [pH 8], 10 mM CaCl₂, 1 μM ZnCl₂, 150 mM NaCl). The gels were stained with Coomassie blue G-250 and destained with water.

Preparation of anti-NprB antibodies.

The protein NrpB was extracted from 2-DE gels of the NBY-O₂ secretome of B. anthracis ΔVollum and purified using the Maxi GeBAflex tube dialysis system (Gene Bio-Application, Israel). Three doses of gel-purified NrpB were administrated to six mice (5 to 10 μg/animal) at 2-week intervals (first injection in complete Freund adjuvant [Sigma]; second and third injections in incomplete Freund adjuvant). Mice were sacrificed 2 weeks after the third administration, and their sera were separated from the hematocrit, pooled, and used as specific anti-NrpB antibodies. The specificity of the sera was confirmed by Western blot analysis. Mice were handled according to the National Institutes of Health's guide for the care and use of laboratory animals and the guidelines of the local commission for animal care.

RESULTS

Experimental design.

B. anthracis secreted proteins were obtained from different cultures of the wild-type virulent Vollum strain (pXO1⁺ pXO2⁺) and ΔVollum plasmid-cured (pXO1⁻ pXO2⁻) bacteria (145). The studies included also the B. anthracis Δ14185 strain (27, 94), a plasmid-cured derivate of the vaccine ATCC 14185 strain that is defective in proteolytic activity (42). The specific mutations conferring the nonproteolytic phenotype of this strain are not known, yet inspection of its extracellular proteome served for determining whether or not the secretome signatures are distorted by the possible effect of bacterial secreted proteolytic activity (e.g., see the effect of the abundant extracellular protease NprB [see below], which is undetected in the Δ14185 cultures).

Proteins secreted by these B. anthracis strains were collected from essentially three types of cultures (Fig. 1): (i) bacteria grown aerobically in the rich media BHI and FAG, conditions which enable optimal protein expression resulting in a rich repertoire of secreted proteins; (ii) bacteria grown in the lower-nutrient-content NBY medium under aerobic conditions (NBY-O₂); and (iii) bacteria grown in NBY medium supplemented with bicarbonate under semianaerobic conditions (NBY-CO₂; see Materials and Methods). The last culture conditions, characterized by a high CO₂ content, were used to mimic those encountered by the bacteria in the host during infection, which are known to induce expression of the B. anthracis toxins and capsule (75). The semianaerobic conditions used for the NBY-CO₂ culture are necessary for preserving a high CO₂ concentration in the culture. It is therefore important to emphasize that the terms NBY-CO₂ and NBY-O₂ pertain to the relative CO₂ concentration rather than to anaerobic or aerobic conditions.

FIG. 1.

Overview of the 2-DE maps of the various B. anthracis secretomes inspected in this study. Representative gels of proteins secreted by different virulent and nonvirulent B. anthracis strains under the indicated growth conditions are shown. All secretomes were collected at the stationary phase of the bacterial cultures (20 h) (see Materials and Methods), and equivalent amounts of protein were loaded on gels.

Composition of the B. anthracis secretome.

The protein signatures of the three different strains under the various conditions (Fig. 1) were obtained by 2-DE fractionation of equivalent proteins amounts. Coomassie blue-stained gels (as exemplified in Fig. 2 and 3) were scanned for quantification of the various protein spots. More than 500 protein spots could be visualized on the various 2-DE gels. The identities of the proteins represented by the spots were determined by MALDI-TOF analysis of their tryptic digestion products (Table 1; see Table SI in the supplemental material). In some cases the identities of different spots could be unequivocally established by their positions on the proteomic maps; in most of the cases, the identity of a certain spot was confirmed by MALDI-TOF analysis. Overall, approximately 400 differently migrating protein spots from the various gels were inspected by MALDI-TOF analysis. Under the various conditions used in this study, the major proteins constituting the B. anthracis secretomes represent the products of 64 ORFs (Table 1). Twenty-two proteins have not been previously detected in the B. anthracis secretome and are marked “novel” in Table 1. The vast majority of the proteins detected in this study are chromosomally encoded, regardless of the presence or absence of the virulence plasmids. In fact, with the exceptions of PA, LF, and EF, only three pXO1-encoded proteins were identified: pXO1-15, pXO1-90, and pXO1-130. Remarkably, no pXO2-encoded proteins were detected in the secretome. Nevertheless, the characteristic pXO2-encoded capsule of B. anthracis was detected under conditions of high CO₂ tension (not shown).

FIG. 2.

2-DE separation of proteins secreted by B. anthracis strain Vollum cultured in rich FAG and BHI media. The numbers in the left lower corners of the panels correspond to those in Fig. 1. For a detailed identification of the various spots marked on the gels, see Table 1. Spots within boxes are isoforms of same protein. MW, molecular weight (in thousands); NL, nonlinear.

FIG. 3.

2-DE separation of proteins secreted by B. anthracis strain Vollum cultured in NBY medium. Representative 2-DE gels of the NBY-O₂ and NBY-CO₂ medium secretomes of the virulent B. anthracis Vollum strain are shown. The numbers in the left lower corners of the panels correspond to those depicted in Fig. 1. For a detailed identification of the various spots marked on the gels, see Table 1. All spots within boxes represent isoforms of same protein. White boxes indicate full-length or partial fragments of PA. Arrows indicate the major proteins InhA1 (spot 2), NprB (spot 46), and HtrA (spot 11) (see text). MW, molecular weight (in thousands); NL, nonlinear.

TABLE 1.

B. anthracis proteins identified in this study

Spot no.a	Accession no. (TIGR)	Functional annotation and designationb	Function category	Secretion signals, retention sequences, and/or functional domainsc	Occurrenced in:									Previously detected in B. anthracis proteomic surveys (reference[s])e
					FAG/BHI			NBY-O2			NBY-CO2
					Δ14185	ΔVollum	Vollum	Δ14185	ΔVollum	Vollum	Δ14185	ΔVollum	Vollum
1	BA4539	Chaperone, heat shock protein; DnaK	Protein folding; stress response			•	•				•		•	6, 25, 41, 50, 64, 85
2	BA1295	Metalloprotease; InhA1 (Bt)	Protein degradation	S	•	•	•	•	•	•	•	•	•	6, 41, 50, 64
3f	BA3367	γ phage receptor; GamR	Unknown	S, LPXTG	•	•	•				•	•	•	6
4	BA0267	Chaperone, heat shock protein; GroEL	Protein folding; stress response		•	•	•				•	•	•	6, 25, 41, 50, 64
5	BA5364	Enolase; Eno	Metabolism (glycolisis)		•	•	•				•	•	•	6, 41, 50, 78
6	PXO1-110	Protective antigen; PA	Toxin	S			•						•	78
7	BA0685	Unknown	Unknown	S, 3D	•	•	•				•	•	•	6, 50
8g	BA1191	SBP of ABC transporter (oligopeptide)	Transport	S	•	•	•				•	•	•	6, 41, 50, 78
8g	BA0656	SBP of ABC transporter (oligopeptide)	Transport	S	•	•	•				•	•	•	6, 41
9	BA4322	Nucleotidase	Nucleotide and nucleoside interconversions	S	•	•	•	•				•	•	6
10	BA5470	Sulfatase; YvgJ (Bl)	Cell envelope biogenesis	S, 5× TM	•	•	•				•	•	•	6, 50, 78
11	BA3660	Serine protease; HtrA	Protein folding/secretion; stress response	S	•	•	•				•	•	•	Novel (41)
17	BA3588	Lipoprotein; VanW	Cell envelope	S	•	•	•						•	6, 50, 78
18	BA3189	SBP of ABC transporter (Mn); MntA	Transport	S		•	•					•	•	Novel
19	BA0885	S-layer protein; Sap	Cell envelope	S, SLH	•	•	•				•	•	•	6, 25, 41, 50, 78
20	BA1197	SBP of ABC transporter (oligopeptide)	Transport	S	•	•	•				•	•	•	6, 25, 64, 85
21	BA1290	Camelysin (Bc); TasA (Bs)	Protein degradation	S	•	•	•						•	6, 64
22	BA0703	Quinol oxidase subunit II; QoxA (Bl)	Energy metabolism; electron transport	S, 2× TM	•	•	•						•	Novel (41)
23	BA0228	ATP-binding subunit of ABC transporter; YdiF (Bl)	Transport		•	•	•	•	•	•		•	•	Novel
27h	BAS5205	Collagen adhesion protein	Cell adhesion	S, VG	•	•	•							Novel
28	BA0908	SBP of ABC transporter (oligopeptide)	Transport	S	•	•	•	•				•		Novel (41)
29	BA2947	Sulfatase; YflE (Bl)	Cell envelope biogenesis	S, 5× TM	•	•	•	•				•		6, 50
30	BA0855	SBP of ABC transporter (amino acid); YckK (Bl)	Transporter	S	•	•	•							Novel (41)
31	BA4750	d-Alanyl-d-alanine carboxypeptidase; VanY	Cell envelope (degradation of peptidoglycan)	S	•	•	•							6, 50
32	BA2230	Unknown; YpjP (Bl)	Unknown	S	•	•	•							6
33	BA3346	6-Aminohexanoate dimer hydrolase	Cell wall remodeling	S, β-lactamase	•	•	•							Novel
34	BA2944	Polysaccharide deacetylase; YjeA (Bl)	Degradation of polysaccharides	S	•	•	•						•	6
36	BA0165	Prolyloligopeptidase	Protein degradation	S	•	•	•					•		Novel
37	BA3560	Phosphodiesterase	Degradation of phospholipids and lipoproteins	S	•	•	•							Novel (41)
38	BA1952	NlpC/P60 family endopeptidase	Cell wall hydrolysis	S, SH3 ×2, NlpC/P60	•	•	•					•	•	6
39	BA0799	Unknown	Unknown	S, HlyD	•	•	•						•	6, 41
40	BA0331	Polysaccharide deacetylase; YxkH (Bl)	Biosynthesis/modification of peptidoglycan	S	•	•	•							6, 50
41	BA2793	Chitinebinding protein	Degradation of polysaccharides	S, FN3	•	•	•						•	Novel (41)
42	BA1449	Peptidase M23/M37	Degradation of proteins	S	•	•	•					•		6
43	BA3162	Nucleotidase	Nucleotide and nucleoside interconversions	S	•	•	•					•		6
45	pXO1-15	Unknown	Unknown							•				Novel (41)
46	BA0599	Neutral protease; NprB (Bs)	Protein degradation	S					•	•				6, 78
47	BA0887	S-layer protein; EA1	Cell envelope	S, SLH	•	•	•					•		6, 25, 50, 64, 78
48	BA3338	Unknown	Unknown	S, SLH	•						•	•		Novel (25, 41, 85)
49	BA3737	Alanine amidase II; CwlA	Cell wall hydrolysis	S, SLH	•	•	•							6, 25, 41, 50
50	BA0898	Alanine amidase III; CwlB	Cell wall hydrolysis	S, SLH	•	•	•						•	6, 25, 50
51	PXO1-90	Unknown	Unknown	S, SLH									•	41, 78
52	PXO1-107	Lethal factor; LF	Toxin	S									•	78
53	BA1436	Sulfatase YvgJ (Bl)	Cell envelope biogenesis	S, 5× TM	•	•	•							50
54	BA1075	Exo/endonuclease/phosphatase	Nucleic acid degradation	S	•	•	•							Novel (41)
55	BA2283	Unknown	Unknown		•	•	•							Novel
56	BA0796	Unknown	Unknown	S, SH3 ×2, 3D	•	•	•		•	•		•	•	6, 41
57	BA4346	2,3-Cyclic nucleotide 2-phosphodiesterase; CpdB or YfkN (Bl)	Nucleic acid degradation	S, LPXG	•	•	•				•			6, 50
58	BA2460	Unknown; FenI (Bc)	Unknown		•	•	•	•	•	•				Novel
59	BA5427	Endopeptidase; LytE	Cell wall hydrolysis	S, NlpC/P60 ERM		•	•						•	6, 41
60	PXO1-122	Edema factor; EF	Toxin	S			•						•	78
61	BA0307	Lipoprotein of unknown function; YerH (Bl)	Unknown	S	•	•	•						•	Novel
62	BA2673	Chitosanase	Degradation of polysaccharides		•	•	•							Novel
63	BA3854	Exochitinase; Chl36	Degradation of polysaccharides		•	•	•	•	•	•				Novel
64	BA5696	Superoxide dismutase cofactored by Mn; SodA-2	Detoxification, oxidative stress response		•	•	•		•	•				6, 25, 41, 50
65	BA5220	SBP of ABC transporter (methionine)	Transport	S	•	•	•							Novel (25, 41)
66	BA2827	Chitin binding protein	Degradation of polysaccharides	S		•			•	•				6, 41
67	BA0345	Alkyl hydroperoxide reductase, subunitC; AhpC	Detoxification, oxidative stress response		•	•	•							25, 41, 50, 78
68	BA4705	Trigger factor; TIG	Protein folding		•	•	•							6, 25, 41, 50
69	BA3645	SBP of ABC transporter (oligopeptide)	Transport	S	•	•	•							Novel (25)
70	BA4387	Leucine dehydrogenase; LeuDH	Energy metabolism, amino acids		•	•	•							41, 50
71	BA4873	Alanine dehydrogenase; Ald-2	Energy metabolism, amino acids		•	•	•							41, 50, 78
72	BA5249	3-Hydroxyacyl coenzyme A dehydrogenase	Fatty acid and phospholipid degradation	S	•	•	•							Novel (41)
73	pXO1-130	SBP of ABC transporter (Zn); AdcA	Transport	S			•						•	78

^aEach entry represents a distinct protein. Spots representing subfragments of proteins which were also detected as full-length polypeptides (such as spots 12 to 16 and 26, representing subfragments of PA, in Fig. 3) are not included. The entries describing the known components of the B. anthracis toxin PA, LF, and EF (spots 6, 52, and 60, respectively) are in boldface.

^bWhen available, the designations (short names) used for the various proteins are included. In some cases, the short names are inferred from those used for the respective orthologues from other Bacillus strains: Bs, Bacillus subtilis; Bl, Bacillus licheniformis; Bc, Bacillus cereus; Bt, Bacillus thuringiensis.

^cDomains which are inherently related to the functional annotation of the proteins are not indicated. S, export signal peptide; SLH, S-layer homology domain; LPXTG, gram-positive sortase domain; TM, transmembranal domain; SH3, Src homology domain; VG, Pneumoviridea glycoprotein G domain; FN3, fibronectin type III domain; HlyD, homology domain to proteins involved in toxin secretion in gram-negative bacteria; 3D, cation-binding domain; NLP/P60, cell wall peptidase family domain; ERM, ezrin/radixin/moesin family domain.

^dThe secretomes in which the various proteins were detected are indicated by a dot. The relative abundances of the respective proteins in the various secretomes are depicted in Fig. 5. Blanks indicate that the protein represents less than 0.1% of the total protein present in the secretome analyzed.

^eB. anthracis proteins reported for the first time in the B. anthracis secretome are marked as novel. Eight previously reported proteomic surveys were scrutinized; three of them focused on the composition of B. anthracis secretome (6, 50, 78), while five other studies (25, 41, 64, 85, 144) focused on cell-associated proteins; proteins identified in one of these five studies only are still marked novel.

^fThe LPXTG domain protein BA3367 (spot 3), which is annotated as functionally unknown in the TIGR B. anthracis genomic database, has recently been shown to represent the γ phage receptor of B. anthracis (31).

^gThe proteins BA1191 and BA0656 (both of which are solute-binding subunits of ABC transporters) were consistently identified in the same protein spot; therefore, they are both marked as spot 8. The predicted ligand specificities of all SBP of ABC transporters are provided in parentheses.

^hSpot 27 was identified as a collagen adhesion protein present in the B. anthracis Sterne, Vollum, and Ames strain protein databases. This protein is inadvertently absent from the B. anthracis Ames ancestor protein database (used for protein identification in this study) despite a bona fide putative ORF being present at the correct locus in the genomic DNA sequence. Therefore, the accession number provided is that of the Sterne protein.

At least 31 proteins identified in the secretome are strongly related to virulence of other pathogenic bacteria (Table 2). This observation suggests that B. anthracis secretes, in addition to its “classic” lethal toxin and edema toxin, a large repertoire of proteins which may be essential for its virulence.

TABLE 2.

Secreted proteins with documented involvement in bacterial pathogenesis detected in this study^a

Spot no.b	Accession no.	Protein	Documented involvement in virulencec (reference[s])
1	BA4539	DnaK	Chlamydia trachomatis, immunogenic (124); Listeria monocytogenes (56, 125); Vibrio cholerae (22); Salmonella enterica, required for invasion and survival in MΦ (19, 132)
2	BA1295	InhA1	Bacillus thuringiensis, required for MΦ escape (119); Bacillus cereus, exosporium protein (23)
4	BA0267	GroEL	Chlamydia trachomatis, vaccine candidate (124); Listeria monocytogenes (43); Salmonella enterica and others (19)
5	BA5364	Eno	Streptococci, involved in tissue invasion, immunosuppressive (32, 141); Candida albicans, immunogenic, protective, vaccine candidate (39); Listeria monocytogenes (125)
6	pXO1-110	PA	Bacillus anthracis toxin
18	BA3189	MntAd	Streptococci and other (13, 132, 133); B. anthracis (48)
73	pXO1-130	AdcAd	Brucella abortus, required for intracellular survival (73); Streptococcus pneumoniae (33); Haemophilus ducreyi (45, 83)
8	BA1191, BA0656	Opp familyd	Bacillus thuringiensis: (37, 51, 128); Listeria monocytogenes, required for intracellular survival (16); Streptococcus pyogenes (130)
20	BA1197	Opp familyd
28	BA0908	Opp familyd
69	BA3645	Opp familyd
9	BA4322	Nucleotidase	Pseudomonas aureginosa, cytotoxic in MΦ (157); Vibrio cholerae, cytotoxic in MΦ (118); Haemophilus influenzae, immunodominant, vaccine candidate (158)
43	BA3162	Nucleotidase
10	BA5470	Sulfatase	Pseudomonas aureginosa, Helicobacter pylori, Burkholderia cepacia (57); Eschericia coli (62); Clostridium difficile (126); Mycobacterium tuberculosis (104); Campylobacter jejuni (155)
29	BA2947	Sulfatase
53	BA1436	Sulfatase
11	BA3660	HtrA	Pasteurella piscicida (61); Chlamydia pneumoniae (101); Enterotoxigenic E. coli (138); Helicobacter pylori (55); Haemophilus influenzae (86); Yersinia enterocolitica (60, 84); Salmonella typhimurium (44, 68); Streptococcus pneumoniae (66); Klebsiella pneumoniae (28); Streptococcus pyogenes (55, 69)
21	BA1290	Camelysin	Bacillus cereus (52)
27	BAS5205	Collagen adhesin	Staphylococcus aureus (106, 150, 151); Pneumoviridea virulence (88)
34	BA2944	Polysaccharide deacetylase	Streptococcus pneumoniae (14, 143)
40	BA0331	Polysaccharide deacetylase
36	BA0165	Prolyl olygopeptidase	Streptococcus pneumoniae (113); Salmonella enterica (102); Porphyromonas gingivalis (10); Trypanosoma cruzi (53)
38	BA1952	NlpC/P60 family protein	Listeria monocytogenes (92); Staphylococcus aureus, Clostridium acetobutylicum, Mycobacterium tuberculosis (3); Corynebacterium diphtheriae (149)
41	BA2793	Chitin-binding protein	Staphylococcus areus (54); Listeria monocytogenes (34)
46	BA0599	NprB	Listeria monocytogenes (26); Bacillus cereus (110)
49	BA3737	Alanine amidase	Clostridium, Enterococcus faecalis, Pseudomonas aureginosa, Neisseria meningitidis, Streptococcus agalactiae, Staphylococcus, Yersinia pestis (122); Salmonella enterica (40); Listeria monocytogenes (21); Bacillus anthracis, immunodominant membranal proteins (8, 25)
50	BA0898	Alanine amidase
52	pXO1-107	LF		Bacillus anthracis toxin
57	BA4346	YfkN		Clostridium perfringes (11); Yersinia enterocolitica (156)
59	BA5427	LytE	Listeria monocytogenes (92); Staphylococcus aureus, Clostridium acetobutylicum, Mycobacterium tuberculosis (3); Corynebacterium diphtheriae (149); ERM domain is involved in adhesion (90, 127)
60	pXO1-122	EF	B. anthracis toxin
64	BA5696	Superoxide dismutase, Mn	Listeria monocytogenes, highly immunogenic and vaccine candidate (59); Mycobacterium tuberculosis (18); Vibrio vulnificus (72); Brucella abortus (112)
67	BA0345	AhpC	Salmonella enterica, immunogenic (134); Mycobacterium tuberculosis (129); Helicobacter pylori, immunogenic (111, 154); Legionella pneumophilla, required for survival in MΦ (120); Mycobacterium leprae, required for survival in MΦ (115); Bacillus anthracis, immunodominant membranal protein (25)

^aThe table represents a partial list of known and potential virulence factors among B. anthracis secreted proteins.

^bProteins are listed according to their respective spot numbers, except for the paralogs identified in this study (three sulfatases [spots 10, 29, and 53], two nucleotidases [spots 9 and 43], two alanine amidases [spots 49 and 50], and two polysaccharide deacetylases [spots 34 and 40]), which are grouped. Also grouped are proteins representing solute-binding subunits of ABC transporters. The B. anthracis classic toxin components are in boldface.

^cAll of the selected pathogenic bacterial systems and references provided exemplify the involvement of these proteins in virulence. In some cases additional brief information pertaining to the role and significance of the proteins is also provided. MΦ, macrophage.

^dABC transporter (solute-binding subunit).

Secretion signal peptides and cell wall retention sequences in proteins identified in the secretome. (i) Signal peptides.

Fifty proteins in the secretome exhibit export signal peptides (Table 1). The signal sequences displayed by the B. anthracis secreted proteins are highly similar to those described for Bacillus subtilis, in which the secretion process and secretome have been extensively studied (4, 5, 70, 135, 136, 139). All 50 signals conform to the consensus sequences of type I or type II SPases, and the proteins should therefore be secreted via the Sec pathway. The lengths of most of the predicted signals vary between 20 and 41 amino acids. The protein HtrA (spot 11) exhibits an unusually long secretion signal (58 amino acids). As expected, most of the signal peptides of the putative lipoproteins (processed by type II SPases) are shorter (19 to 21 amino acids) and invariably include a typical cysteine residue at the +1 position. Fourteen proteins identified in this study do not exhibit a recognizable signal sequence. This may be due to the fact that some secretion signals do not conform to the consensus sequence, and the presence of such proteins in bacterial secretomes may indicate the existence of secretion pathways other than those known to involve a canonical export signal. The possibility that these proteins are cytosolic in nature and their appearance in the secretome results from cell lysis cannot be ruled out. Yet, we note that orthologs of eight proteins which do not possess a recognizable signal sequence (DnaK, GroEL, enolase, SodA-2, AhpC, TiG, chitosanase, and exochitinase Chi36) have been detected in an extracellular form in previous studies of many other bacteria (18, 50, 80, 107).

The 2-DE electrophoretic migrations of several proteins indicated putative removal of extensive segments from the full-length forms, more than would be expected from processing of regular signal sequences. These proteins are as follows: some PA subforms (Fig. 3); the 3 sulfatases BA5470 (spot 10), BA2947 (spot 29), and BA1436 (spot 53); an InhA1 subform (spot 2) (Fig. 3); the protein QoxA (spot 22); the protease NprB (spot 46); and the S-layer homology (SLH) protein BA0898 (spot 50). In all cases, examination of the respective MALDI-TOF spectra established the approximate boundaries of these truncations (see below). In the case of the three sulfatases mentioned above (Fig. 4), an explanation for the occurrence of truncated forms may originate from the fact that these proteins exhibit a signal-like sequence located approximately 180 amino acids from their N termini. This phenomenon of an internally located signal peptide was reported by Antelman and coworkers (6) for two of the proteins (BA5470 and BA2947). We note that each of the three sulfatases exhibits extensive transmembrane domains (five hydrophobic loops) located within the N-terminal portion of the molecule (Fig. 4; see below), the removal of which could convert these proteins into the soluble form, compatible with their occurrence in the secretome. A mechanism which ensures the dual location of these sulfatases, as both membrane-bound and secreted forms, may be beneficial to the bacteria in invasiveness and pathogenic colonization (57, 62, 104, 126, 155).

FIG. 4.

Evidence for removal of transmembrane domains from three secreted B. anthracis sulfatases identified in this study. The positions of the electrophoretic migrations of the three sulfatases BA2947, BA1436, and BA5470 are indicated on the 2-DE gel (enlarged portion of gel 1 from Fig. 1) depicted in the left panel, and their Kyte-Doolittle hydropathy profiles (76) are provided in the right panel. The secretion signal-like sequence is indicated; the predicted SPase I cleavage site is indicated by a dashed line. Shaded boxes under the hydropathy profiles indicate the positions of the tryptic peptides identified in the MALDI-TOF spectra.

(ii) Membrane retention sequences.

The occurrence of typically cell-associated proteins in the extracellular milieu is an intriguing phenomenon, which has been documented in B. subtilis and other bacteria (4, 107, 137, 142). The exact mechanism of the release of membrane-anchored or covalently cell-wall-attached proteins is still not resolved. In the present study, 22 proteins detected in the B. anthracis secretome possess retention sequences (Table 1): 4 proteins exhibit hydrophobic transmembrane segments (the three sulfatases described above [BA5470/spot 10, BA2947/spot 29, and BA1436/spot 53] and the protein QoxA/spot 20), 2 proteins exhibit an LPXTG sortase recognition sequence (93) for covalent anchoring (BA3367/spot 3 and BA4346/spot 57), 9 lipoproteins representing the ligand-binding subunits of ABC transporters exhibit a lipobox for lipoprotein covalent attachment (BA1191 and BA0656/spot 8, BA3189/spot 18, BA1197/spot 20, BA0908/spot 28, BA0855/spot 30, BA5220/spot 65, BA3645/spot 69, and pXO1-130/spot 73), and 6 proteins possess SLH domains (Sap/spot 19, EA1/spot 47, BA3338/spot 48, BA3737/spot 49, BA0898/spot 50, and pXO1-90/spot 51).

Ligand-binding subunits of ABC transporters have been reported to represent a major class of secreted proteins in other gram-positive bacteria (4, 137). We note that in previous proteomic analyses of B. anthracis membranes (25), only two of the nine ABC transporter solute-binding subunits detected in the present study were identified: BA5220 (spot 65) and BA3189 (MntA, spot 18). MntA has recently been shown to represent an important virulence determinant of B. anthracis (48).

The S-layer proteins Sap and EA1 are known to exist also as subunits in the extracellular medium, and Sap has been shown to be released in considerable amounts (95, 97). Interestingly, all five chromosomally encoded SLH domain-harboring proteins observed in the secretome were detected in the B. anthracis membrane fraction in a previous proteomic study (25).

Functional classification of the secreted proteins.

Twelve proteins identified in this study can be defined as proteins with an unknown function; their frequency (18%) is somewhat lower than their representation in the genome (30%) (7, 8, 121), a phenomenon often encountered in proteomic studies (see reference 25 for a discussion addressing this issue). Furthermore, we note that 7 of the 12 proteins of unknown function exhibit recognizable domains: BA0685 (spot 7) possesses a 3D domain (putative peptidase domain), BA0796 (spot 56) exhibits both a 3D domain and an SH3 domain (involved in protein-protein interactions), BA0799 (spot 39) exhibits a truncated Hly domain (involved in toxin secretion in gram-negative bacteria), BA3338 (spot 48) and pXO1-90 (spot 51) exhibit SLH anchorage domains, the lipoprotein BA3588 (spot 17) is a putative surface-anchored protein, and BA3367 (spot 3) exhibits a sortase recognition LPXTG motif. This ORF of unknown function has been recently recognized as a γ phage receptor (31).

Almost 80% of the polypeptides identified are functionally annotated (see Table 1 for a short description of their function), allowing their classification into six functional categories (Table 3): enzymes, ABC transporters, chaperons, proteins active in the process of detoxification, structural proteins, and toxins (the last category refers to the classical anthrax toxins PA, LF, and EF).

TABLE 3.

Functional classification of proteins identified in the B. anthracis secretome^a

Category	Function	No. of proteins	Spot nos.
Enzymes	Sulfatase	3	10, 29, 53
	Chitinase	4	41, 62, 63, 66
	Polysaccharide degradation	2	34, 40
	Nucleotidase	4	9, 43, 54, 57
	Phospholypid degradation	2	37, 72
	Protein/peptidoglycan/amino acid degradation/modification	16	2, 11, 21, 27, 31, 33, 36, 38, 42, 46, 49, 50, 52, 59, 70, 71
	Energy metabolism	4	5, 22, 70, 71
ABC transporters	Transport of amino acids or peptides	7	8, 20, 28, 30, 65, 69
	Transport of metal ions	2	18, 73
	ATP-binding subunit of ABC transporter	1	23
Chaperonins, protein folding		4	1, 4, 11, 68
Stress (detoxification)		4	1, 4, 64, 67
Structural	S-layer protein	2	19, 47
Toxins	PA, LF, EF	3	6, 52, 60
Unknown function	Exhibit recognizable domains	7	3, 7, 17, 39, 48, 51, 56
	No recognizable domains	5	32, 45, 55, 58, 61

^aSee Table 1 for a detailed description of the functions of individual proteins. Some proteins have a dual role and therefore are included in two functional categories: GroEL (spot 4) and DnaK (spot 1) are categorized as chaperones, yet these are heat shock proteins induced under stress conditions; HtrA (spot 11) is a protease and a chaperone involved in the response to secretion stress; LeuDH and Ald-2 (spots 70 and 71, respectively) belong both to the energy metabolism category and the amino acid degradation group; and the toxin component LF is a metal protease and therefore is included both in the toxins and the protein degradation enzymes categories. Spot 8 represents two proteins (see Table 1); therefore, the total number of proteins in the category “transport of amino acids or peptides” is 7.

The numbers of secreted proteins attributed to the various functional categories (Table 3) reveal that B. anthracis releases into the medium an unusually high number of proteins possessing activities involved in polypeptide processing: 16 enzymes have proteolytic and amino acid-degrading activity, and 7 transporters exhibit specificity for oligopeptides or amino acids. This characteristic is even more pronounced when one considers the abundance rather than the number of polypeptides possessing protein hydrolysis functions (between 20 and 95% of the total protein content visualized in the 2-DE gels) (Fig. 5; see below).

FIG. 5.

Relative abundances of B. anthracis proteins in various secretomes. The values represent the averaged abundance (±10%) obtained for individual spots identified in at least three 2-DE duplicate gels representing each protein signature, which were scanned and analyzed using the PDQuest software. The absence of spots indicates an abundance lower than 0.1% of the total protein mass. Equivalent amounts (100 μg) of protein were loaded on each gel. Note that some protein spots (e.g., spots 23 and 56) are present in similar abundances in all secretomes and could be used for quantitative comparisons. Gray bars represent proteins possessing biological activities associated with utilization of proteins such as proteases and transporters involved in amino acid and peptide import.

With the exception of two polypeptides (InhA1 and NprB, which are orthologs of PlcR target genes [see below]), none of the proteins identified in the present study are potential members of the PlcR regulon, which governs expression of virulence factors in B. thuringiensis and B. cereus (encoded by orthologs of target genes of the PlcR factor [67] or by genes exhibiting PlcR-inducible cis-acting recognition sequences upstream of their coding sequences [121]). The absence of the PlcR putative targets is in line with the known inactivation of the PlcR regulator in B. anthracis (2, 96).

Differential expression of secreted proteins.

The main general findings following inspection of the various secreted-protein signatures (presented schematically in Fig. 5) are as follows: (i) as expected, a much larger protein repertoire is secreted in rich medium than in minimal medium; (ii) in minimal aerobic medium, more than 90% of the total secreted polypeptides are represented by two proteins; (iii) there are very minor differences between the secretomes of the virulent Vollum and avirulent plasmid-cured ΔVollum strains generated in rich medium or minimal medium under aerobic conditions; (iv) the presence of the virulence plasmids exerts a very significant impact on the composition of the secretome when the cells are cultured in minimal medium under high CO₂ tension, with respect to both chromosome- and plasmid-encoded proteins; and (v) the expression of the major extracellular proteases of B. anthracis is influenced by CO₂ and/or by the presence of the virulence plasmids. These observations are detailed below (see also Table 1 and Fig. 5).

(i) Rich-medium secretomes.

Approximately 500 protein spots in a wide dynamic range of concentrations can be visualized by Coomassie blue staining of the proteomic maps of the rich-medium secretomes (Fig. 1 and 2). Small amounts of PA and EF could be detected in the secretome of the Vollum strain, suggesting that their repression in the absence of CO₂ (75) is leaky. Except for these pXO1-related features, the Vollum and ΔVollum rich-medium secretomes are strikingly similar and differ only slightly in terms of the abundances of some of the proteins (Fig. 5). Thus, at least in the case of secreted proteins, possible regulatory cross talk mechanisms connecting plasmid- and chromosome-encoded genes are not manifested under these highly permissive conditions.

(ii) NBY-O₂ medium secretomes.

Under NBY-O₂ medium growth conditions, the number of secreted proteins is dramatically reduced compared to that of the rich-medium secretomes (Fig. 3 and 5). The secretomes of both the Vollum and ΔVollum strains display unique and rather unexpected compositions, consisting of only two major proteins which together represent more than 90% of the total proteins released by the bacteria: immune inhibitor A (InhA1, BA1295) and neutral protease (NprB, BA0599). We note that a few additional minor chromosome-encoded proteins (Fig. 5) are present in the NBY-O₂ secretomes of both strains (spots 23, 56, 58, 63, 64, and 66). The abundance of these additional protein spots in NBY-O₂ is similar to that observed in the rich medium, further emphasizing the significant induction exhibited by InhA1 and NprB. Both InhA1 and NprB are metalloprotease homologs of virulence factors in the phylogenetically related pathogens B. cereus and B. thuringiensis (24, 49, 50, 110). In both Vollum and ΔVollum, the InhA1 protease appears in two forms that differ in molecular weight (Fig. 3, spot 2). The isoform of lower mass (60 kDa) lacks the C-terminal region of the molecule. The electrophoretic migration of NprB (Fig. 3, spot 46) indicates an approximate molecular mass of 45 kDa, indicative of the removal of about 180 amino acids from its N terminus. A similar processing step was reported to occur in a variety of bacterial virulence-related secreted metalloproteases, such as thermolysin family proteases from Legionella pneumophila, Pseudomonas aeruginosa, Vibrio cholerae, and Vibrio vulnificus (103).

A pXO1-encoded protein of unknown function, pXO1-15 (spot 45), can be detected in significant amounts in the NBY-O₂ secretome of the Vollum strain. Except for pXO1-15, as in the case with rich medium, essentially no differences are observed between the Vollum and the ΔVollum strains.

(iii) NBY-CO₂ medium secretomes.

The pattern described for the NBY-O₂ secretomes is strikingly different when the cells are cultured in the same minimal medium supplemented with bicarbonate (NBY-CO₂) (Fig. 3 and 5). Thirty proteins, which are absent from the NBY-O₂ culture, can now be detected. Furthermore, unlike the situation in rich medium and NBY-O₂, in NBY-CO₂, the protein signature of the Vollum strain differs significantly from that of the plasmid-cured ΔVollum strain, and this difference is characterized by the presence of both plasmid- and chromosome-encoded proteins. Seventeen proteins are present in the NBY-CO₂ secretome of the Vollum strain in quantities significantly higher than in the NBY-CO₂ secretome of the ΔVollum strain, while 13 proteins are more abundant in the NBY-CO₂ secretome of the ΔVollum strain (Fig. 1, 3, and 5; Table 1). These patterns are described in more detail below.

(a) pXO1-encoded proteins abundant in the NBY-CO₂ secretome of the Vollum strain.

As could be expected, PA represents the major protein in the NBY-CO₂ secretome of the Vollum strain (Fig. 3 and 5). Several forms of PA were detected, representing either isoforms with the same mass but different isoelectric points or subfragments resulting from trimming or processing of the full-length molecule. The major form, migrating in the 2-DE gel at a position compatible with a full-length molecule, represents more than 30% of the total protein amount in the secretome (Fig. 3, framed spot 6) and exhibits a wide range of pI. These observations are in agreement with a recent study which suggested that modifications, such as deamidations, may result in a multitude of PA forms that differ in their isoelectric points (159). Several PA subfragments, representing >10% of total protein visualized on the 2-DE gel, could be detected (Fig. 3, white frames marked 12,26 [around 40 kDa] and 13-16 [around 30 kDa]). Inspection of the MALDI-TOF spectra established that the 40-kDa PA forms originate from the C-terminal portion of the molecule, while the 30-kDa PA forms span the N-terminal region of the molecule. The presence of PA subfragments in B. anthracis culture supernatants has been observed in previous studies (78, 146). The other toxin components, LF and EF, were detected in much smaller amounts, in agreement with previous studies (81).

Two additional pXO1-encoded genes were detected in the NBY-CO₂ secretome of the Vollum strain: pXO1-90 (spot 51) and pXO1-130 (spot 73). Both pXO1-90 and pXO1-130 were documented in a previous report from our laboratory to be immunogens expressed in vivo, and they represent potential anthrax vaccine candidates (7). These two proteins were also identified in the supernatant of B. anthracis cultures under high CO₂ tension by Lamonica and coworkers (78). These proteins may be involved in bacterial pathogenesis: detection of the protein pXO1-90 (an SLH-exhibiting protein of unknown function) is in good agreement with the observation that it is transcriptionally activated by the virulence regulator AtxA (17). The protein pXO1-130 represents a truncated version of AdcA, the solute-binding subunit of an orphan ABC transporter exhibiting extensive domain similarity to proteins involved in the import of zinc (30, 58). AdcA homologs were invoked in virulence (Table 2) of other pathogenic bacteria such as Brucella (73), Pasteurella (46), and Haemophilus (83). Notably, the B. anthracis toxin subunit LF is a metalloprotease cofactored by zinc (74). The up-regulation of a Zn import machinery under conditions favoring toxin synthesis may therefore be beneficial for the bacteria.

(b) Chromosomally encoded proteins abundant in the NBY-CO₂ secretome of the Vollum strain.

Twelve chromosomally encoded proteins were detected under these conditions (Table 1 and Fig. 5, spots 1, 11, 17, 21, 22, 34, 38, 39, 41, 50, 59, and 61). Three of these (spots 17, 39, and 61) are proteins of unknown function. The functional classification of the other nine proteins is discussed below.

The protein DnaK (spot 1) is a chaperone, usually involved in heat, salt, and oxidative stress responses; consistently, DnaK could not be detected in the NBY-CO₂ secretome of ΔVollum at all time points inspected (not shown). This result is puzzling, especially in view of the fact that it can be readily detected in the NBY-CO₂ secretome of the related Δ14185 strain (see Fig. S1 in the supplemental material). Yet, we note that in Vibrio cholerae (22) and in Salmonella enterica (132), DnaK was shown to be required for production and secretion of virulence factors; therefore, one may speculate that it is up-regulated in the Vollum strain as an accessory to toxin production.

The protein QoxA (spot 22), which can exist both membrane bound and as a truncated product in the medium (see above), is an essential terminal electron acceptor in the electron transport chain in bacterial respiration metabolism (116). The expression of QoxA observed in the NBY-CO₂ culture of the Vollum strain may suggest that this protein is required for anaerobic growth, as recently documented for group B Streptococcus. Interestingly, quinol oxidase activity was recently implicated in the capacity of these bacteria to grow in the host (153).

Two putative polysaccharide-degrading enzymes were detected: a chitin-binding protein (BA2793/spot 41) and the polysaccharide deacetylase YjeA (BA2944/spot 34). In the aerobic minimal medium culture (NBY-O₂) of B. anthracis, another chitin-binding protein is expressed (BA2827/spot 66). It should be noted that BA2793 but not BA2827 exhibits two copies of the FN3-type extracellular matrix adhesion domains. Such domains were identified in virulence-related proteins of pathogens such as Staphylococcus aureus (54), Clostridium thermocellum (71), and Streptococcus pyogenes (108).

Five protein-degrading enzymes exhibit elevated levels (Fig. 3 and 5): the serine protease HtrA (BA3660/spot 11), two cell wall hydrolases belonging to the NlpC/P60-family (BA1952/spot 38 and LytE-BA5427/spot 59), the SLH domain alanine amidase (BA0898/spot 50), and the camelysin (BA1290/spot 21).

HtrA (spot 11) is the most abundant protein in the NBY-CO₂ secretome of the Vollum strain except for PA (Fig. 3 and 5). HtrA has been extensively studied in the context of secretion, in which HtrA fulfills two roles: it acts as a chaperone responsible for the correct folding of secreted proteins and as a protease responsible for the degradation of misfolded proteins (5, 29, 65, 136). The high abundance of HtrA in the NBY-CO₂ secretome of the Vollum strain may stem from the fact that under these conditions, it is active in supporting the secretion of extremely large amounts of PA. Such a phenomenon was described for another extracellular chaperone, PrsA, which was shown to be necessary for production of PA in a heterologous B. subtilis system (147).

The protein BA1952 (spot 38) is a putative cell wall hydrolase exhibiting an NlpC/P60 domain and also two SH3 (Src homology) virulence-related domains. The protease LytE (spot 59) is a major cell wall hydrolase in B. subtilis (152). It harbors two important functional domains: an NlpC/P60 domain, similar to BA1952, and a truncated N-terminal ERM domain. The ERM domain is basically a eukaryotic protein-protein interaction motif encountered in actin-binding proteins (90) and was suggested to mediate Shigella entry into epithelial cells (127).

The alanine amidase BA0898 (spot 50) is an SLH protein that was previously detected as a membrane-associated protein of B. anthracis and recognized as an immunodominant vaccine candidate (8, 25). While the full-length polypeptide is present in the rich-medium secretome (Fig. 2) and in the membrane proteome (25), the form detected in the NBY-CO₂ secretome of the Vollum strain is a subfragment spanning the C-terminal region of the molecule, devoid of the cell surface anchorage SLH domain. It is interesting to note that an S-layer protein possessing alanine amidase activity is present in considerable amounts in the secretome of the pathogen Clostridium difficile under conditions which favor very high toxin production (105).

(c) Proteins abundant in the NBY-CO₂ secretome of the ΔVollum strain.

The higher calculated relative abundance of some of the proteins in the ΔVollum secretome compared to that of the Vollum strain (for example, spots 2, 3, 4, 5, and 8 in Fig. 5) may be a reflection of the overall lower level of protein species secreted by the ΔVollum strain in NBY-CO₂ medium compared to the wild-type Vollum strain. Yet, five proteins appear to be preferentially secreted by the plasmid-cured ΔVollum strain in the NBY-CO₂ culture, indicating that the presence of the virulence plasmids not only up-regulates but also represses some chromosomally encoded genes, a phenomenon previously observed by transcriptome analysis (17). The proteins which appear in the present study to be down-regulated by the presence of the plasmids (Fig. 5 and Table 1) are the S-layer proteins Sap and EA1 (spots 19 and 47, respectively), a prolyloligopeptidase (BA0165, spot 36), a 5′-nucleotidase (BA3162, spot 43), and an SLH domain-containing protein of unknown function (BA3338, spot 48). The most abundant protein is Sap. This observation is in accordance with a pioneering report of regulatory cross talk which demonstrated the link between Sap expression and the pXO1-encoded PagR and the AtxA virulence regulators (98). In a B. anthracis heterologous production system we have observed a negative linear correlation between high production and release into the medium of recombinant PA and the amount of Sap secreted (47; O. Gat and A. Shafferman, unpublished data), suggesting that the down-regulation of Sap under conditions of high toxin production is beneficial to the bacteria.

The increase in abundance of the prolylpeptidase BA0165, the nucleotidase BA3162, and the SLH protein BA3338 in the ΔVollum secretome could suggest that they are not essential for manifestation of B. anthracis virulence, although proteins belonging to the first two families are documented virulence factors in other pathogens (10, 102, 157, 158).

Tight CO₂-dependent regulation of the protein NprB.

One of the most striking B. anthracis protein signatures observed in this study is related to the proteases InhA1 (spot 2) and NprB (spot 46), which are present in exceptionally large amounts in the NBY-O₂ cultures of both the Vollum and ΔVollum strains (Fig. 3, 5, and 6A). InhA1 represents between 10 and 20% of the total protein in this secretome, while NprB consistently represents between 60 and 80% of the NBY-O₂ secretome. Several observations suggest that the scarcity of other protein spots in the NBY-O₂ secretomes is not associated with the protease activity of NprB: (i) apart from the NBY-O₂ conditions detailed above, a significant induction of NprB and InhA1 can be observed in BHI medium in late stationary phase (30 h), and yet the protein signature of this culture still includes a large number of spots (Fig. 6C); (ii) an NBY-O₂ secretome collected at an earlier time point (10 h) was almost indistinguishable from the one collected after 20 h in culture (Fig. 6A); and (iii) an NBY-O₂ profile with few secreted proteins was also exhibited by B. anthracis Δ14185, a strain which appears to be defective in NprB secretion (Fig. 5 and 6D).

FIG. 6.

Evidence of tight regulation of the expression of the extracellular proteases NprB and InhA1. A, C, and D: 2-DE gels of secretomes for the indicated strains, growth conditions, and harvesting times. Numbers in the lower left corners of some gels indicate the respective protein maps in Fig. 1. Spots representing NprB and InhA1 are boxed (the protein Sod, indicative of a state of stress [12, 63] is indicated in panel C. B: Western blot analysis using anti-NprB antibodies. Ten micrograms of protein of the indicated secretomes (and samples diluted 1:500 in the case of the NBY-O₂ samples) was fractionated by SDS-polyacrylamide gel electrophoresis (upper panel), and transferred to Western blots, probed with anti-NprB antibodies (1:1,000), and visualized by chemiluminescence (ECL). E: Zymograph gels carried out with the indicated NBY secretomes of the Vollum (V) or ΔVollum (Δ) strain, using casein-, gelatin-, or PA-impregnated gels.

InhA1 was detected in all B. anthracis secretomes inspected, albeit in much smaller amounts. In contrast, the neutral protease NprB could not be detected in any other secretome (Table 1; Fig. 5 and 6) except for a late-stationary-phase rich-medium culture (see below). This absence of NprB, and in particular the shutoff of its expression in NBY-CO₂, was quite striking in view of its unusually high expression in the aerobic NBY medium. Western blot analysis using specific anti-NprB antibodies (Fig. 6B) was used to evaluate the extent of its up- and down-regulation. We find that 500-fold dilution of the NBY-O₂ secretome still allowed NprB detection. On the other hand, Western blot analysis performed with anti-NrpB antiserum on one-dimensional gels (Fig. 6B) and on 2-DE gels (not shown) from rich-medium secretome or from NBY-CO₂ culture did not generate an NprB-specific signal. Normalizing for the amount of protein (100 μg/2-DE gel), one could estimate that at least 1,000 times more NprB is present in the NBY-O₂ secretome than under the other conditions tested.

Given the fact that both NprB and InhA1 are proteases, we inspected the proteolytic activity present in the supernatants of the NBY-O₂ and NBY-CO₂ cultures, by use of zymography gels containing casein and gelatin as generic protease substrates (Fig. 6E). We observe that a significantly lower extracellular protease activity is present in the NBY-CO₂ culture. Notably, similar results were obtained by carrying out the zymography tests with PA-impregnated gels, suggesting that this proteolytic activity may interfere with toxin accumulation in the extracellular milieu (Fig. 6E).

Finally, we emphasize that the dramatic decrease in the amount of NprB (and to a lesser extent InhA1) occurs at high CO₂ tension with equal efficiency in the Vollum, ΔVollum, and Δ14185 strains (Fig. 1, 5, and 6). This strongly suggest that the CO₂-responsive mechanism which down-regulates NprB expression (InhA1 only in the case of Δ14185) is chromosomally controlled.

DISCUSSION

In this report we present a proteomic study of B. anthracis which addresses the influence of the presence of the virulence plasmids on the composition of the bacterial secretome in various media and growth conditions. The study expands the current database of B. anthracis secreted proteins by complementing the recent proteomic analyses reported by Antelman and coworkers (6), Gohar and coworkers (50), and Lamonica and coworkers (78) (Table 1), both with respect to the number of protein species identified and, mainly, regarding the expression patterns of the various proteins.

Of the 400 spots inspected, more than 60 proteins were identified (Table 1), corresponding to the most abundant species, many of which are virulence factors in other pathogens (Table 2) and therefore represent potential targets for individual investigation of their role in B. anthracis virulence. It is interesting that the B. anthracis secretome, as evidenced in this and other studies (6, 50, 78), does not resemble that of the closely related B. cereus or B. thuringiensis (50). Based on the genomic similarity of these species and on the known evolutionary silencing of the PlcR regulon in B. anthracis (2, 96; see below) one would expect that the B. anthracis secretome of a strain devoid of plasmids strain should be similar to that of a B. cereus plcR null mutant. Yet these secretomes are different (based on the analysis carried out by Gohar and coworkers [49], which involved the rich-medium secretome of plcR null and wild-type B. cereus cells). This observation emphasizes the phenotypic difference between B. cereus and B. anthracis despite their genotypic similarity and strongly suggests that regulatory circuits other than the PlcR regulon are active in determining secretion in B. anthracis. Furthermore, this indicates that B. anthracis evolved its own set of secreted factors which contribute to its characteristic behavior and that putative virulence factors particularly involved in anthrax pathogenesis are present in the B. anthracis secretome.

The secreted proteins could be classified among several functional categories (Table 3), establishing the prevalence of functions required for protein utilization (e.g., degradation and import), which are overrepresented in the B. anthracis extracellular proteome (6, 121). Fifty out of the 64 proteins reported here possess secretion signal peptides (Table 1) indicative of their export via the Sec pathway. While most of the bacteria possess only one SecA protein, two homologs of the SecA gene were identified in the B. anthracis genome (BA0882 and BA5421 in the “Ames ancestor” genome) (147), suggesting that this bacterium has an alternative secretion pathway which may be responsible for secretion of some of the signal-less proteins found in the secretome (114). The alternative Sec pathway is often involved in the secretion of virulence factors, and it may mediate release of proteins which do not necessarily exhibit a canonical export signal. Notably, in Mycobacterium tuberculosis, two of the signal-less proteins exported by an alternative Sec pathway are SodA and catalase-peroxidase (18), which resemble the B. anthracis SodA-2 and AhpC identified in this study. Similarly, in Listeria monocytogenes, orthologs of the signal-less proteins DnaK, GroEL, and enolase, all of which were identified in the B. anthracis secretome (Tables 1 and 2), are exported via the alternative SecA2 pathway (80). In addition to soluble proteins fully compatible with their localization in the extracellular milieu, the secretome comprises 22 proteins of a membrane nature, either containing extensive transmembrane hydrophobic domains or exhibiting motifs which mediate surface anchoring. This situation was reported also for other bacteria, such as B. cereus, B. licheniformis, B. subtilis, and S. aureus, and it may indicate that dual localization of some proteins, both cell associated and secreted/shed, may be important to their physiological role (4, 50, 107, 137, 142).

In rich medium, the Vollum and the ΔVollum strains of B. anthracis secrete a large number of proteins, and only minor differences between the secretomes of these strains are observed (Fig. 1, 2, and 5). However, inspection of a late-stationary-phase BHI secretome (a stage at which the cells may experience a state of stress) reveals a sharp increase in the levels of the two proteases InhA1 and NprB (Fig. 6C). We note that Antelman and coworkers (6) also observed an increase in expression of NprB in the late stationary phase of the culture of an other plasmid-cured avirulent B. anthracis strain (UM23C1-2). More dramatically, in the NBY aerobic medium (which contains much lower levels of nutrients than the rich BHI medium), unprecedentedly large amounts of these two proteins are observed (Fig. 1 and 3). In NBY supplemented with CO₂, both of these proteases are down-regulated (Fig. 2, 5, and 6B). In particular, the amount of NprB protease, which in the aerobic medium is unusually high (60 to 80% of total protein mass in the secretome), is reduced to a nondetectable level (by at least 3 orders of magnitude) and virtually, disappears from the secretome. The dramatic silencing of the NprB expression in NBY-CO₂ indicates that under conditions in which the B. anthracis toxins are expressed, it is beneficial for the bacteria to reduce its proteolytic activity. This concept is strongly supported by the zymography profiles of the secretomes (Fig. 6D): we observe that a much higher proteolytic activity acting both on casein and gelatin is secreted in NBY-O₂ than in the NBY-CO₂ cultures. Interestingly, this proteolytic activity is highly efficient in a zymography test carried out on gels impregnated with PA, strongly indicating that this high proteolytic activity may be deleterious to the bacterium under conditions when its survival depends on the secretion of the toxins, such as those encountered in the host during infection. The fact that the same phenomenon of a sharp shutoff of NprB expression occurs equally efficiently in the Vollum and ΔVollum strains demonstrates that CO₂-signaling regulatory circuits independent of the pXO1-carried atxA gene or the presence of pXO2 plasmid are active in B. anthracis. This possibility was suggested by Mock and Mignot (100), based on inspection of the CO₂ response of B. anthracis strains devoid of either virulence plasmid, yet the present study may provide direct evidence that chromosomally encoded CO₂-responsive regulons also are involved in determining the virulence of the bacteria.

One of the most fundamental aspects of B. anthracis virulence in contrast to its phylogenetically close relative B. cereus is the evolutionary silencing of the PlcR regulon in B. anthracis. This genetic inactivation is irreversible due to a nonsense mutation in the plcR gene (2), and it was postulated that the manifestation of B. anthracis virulence is strictly dependent on the complete abolishment of the expression of PlcR target genes which may be incompatible with the characteristic pathogenicity of B. anthracis (97). Our results bring strong new evidence in support of this concept; we note that among all secreted proteins identified in this study, the only ones which are homologous to B. cereus potential PlcR targets (67) are the strikingly differentially expressed NprB and InhA1 proteases (although they do not exhibit PlcR-responsive DNA sequences, they are orthologs of PlcR targets). While B. anthracis virulence is not disturbed by PlcR target proteins due to their stable dormant status, both NprB and InhA1, which may be required for survival under minimal medium conditions, are suppressed by a CO₂-dependent mechanism.

In the CO₂-supplemented medium the patterns of expression of many proteins differ in the Vollum and the ΔVollum strains, demonstrating that regulatory cross talk mechanisms, manifested in both an inductive manner and a repressive manner, link the expression of plasmid- and chromosome-encoded proteins. Actually, 10 chromosomally encoded proteins (BA0307, BA0703, BA0799, BA0898, BA1290, BA2793, BA2944, BA3588, BA4539, and BA5427) which were identified in the NBY-CO₂ secretome of the Vollum strain could not be detected at all under the same conditions in the secretome of the ΔVollum strain, suggesting that these gene products are positively regulated by plasmid-encoded factors. On the other hand, expression of 14 proteins (BA0165, BA0267, BA0656, BA0885, BA0887, BA0908, BA1191, BA1295, BA1449, BA2947, BA3162, BA3338, BA3367, and BA5364) encoded by chromosomal genes appears to be down-regulated by the presence of the virulence plasmids. Two additional proteins, BA1952 and BA3660, are present in much larger amounts in the NBY-CO₂ secretome of the Vollum strain, yet they can also be detected in ΔVollum. These two proteins are actually proteases, which appear to be the most abundant proteins of the NBY-CO₂ secretome of the virulent Vollum strain (except for PA) (Fig. 5) and are absent from the NBY-O₂ culture. BA1952 exhibits an NlpC/P60 domain which was observed in peptidases with a widespread role in the dynamics of the bacterial cell wall (3). Most interestingly, BA1952 contains also two SH3 (Src homology) domains. These domains represent bacterial versions of eukaryotic motifs encountered in many signal transduction proteins (38). Furthermore, bacterial proteins possessing SH3 domains were shown to be involved in the virulence of several pathogens, such as the diphtheria toxin repressor DtxR of Corynebacterium diphtheriae (89, 149), the invasion protein InlB of Listeria monocytogenes (92), and the iron-dependent repressor IdeR of Mycobacterium tuberculosis (89) (Table 3). The second abundant protein, BA3660, is HtrA, a chaperone and protease and a well-established virulence factor and potential vaccine candidate in many bacteria, which also has a role in the secretion process (Table 3). Whether HtrA is only an accessory to toxin secretion or is by itself involved in virulence manifestations, such as proteolysis of host tissues, is still to be determined, and studies which address this issue are currently being carried out in our laboratory.

In conclusion, the study presented here strongly suggests that cross talk regulatory circuits affecting expression of chromosome- and plasmid-encoded genes act in a CO₂-dependent fashion. Overall, in comparison to bacteria cured of the virulence plasmids, the secretome composition of the virulent isogenic B. anthracis strain propagated in NBY-CO₂ (conditions which simulate those encountered in vivo) shows that the presence of the plasmids preferentially affects the expression of at least 26 chromosomal genes besides pXO1-carried genes. It is noteworthy that at least 17 of these 26 differentially expressed proteins were invoked in virulence of other pathogens. The pattern of expression characterizing the major secreted proteases identified in this study, e.g., HtrA and NprB, delineates two alternative but not mutually exclusive regulatory circuits: HtrA represents a class of proteins which are subject to positive regulation by factors encoded on the virulence plasmids acting in a CO₂-dependent fashion, and NprB belongs to a category of proteins which are negatively regulated also in a CO₂-dependent modality but by chromosomally encoded factors.